#!/usr/bin/env bash
# Pre-upload artifact hygiene (spec §25): scan every staged artifact for the
# run's secret values and for generic credential patterns, and fail the upload
# on any hit. Secrets must never reach console logs, evidence files, prompts, or
# uploaded artifacts.
#
# Matching is deliberately broader than a literal grep: log formatters wrap long
# values across lines, and encoders re-shape them, so each artifact is also
# scanned in a whitespace-stripped form and against derived encodings of each
# secret. Archives are refused rather than scanned opaquely.
#
# Reads: ARTIFACTS_DIR, plus whichever secrets are in scope for this job. At
# least one of GH_TOKEN / FIREWORKS_API_KEY must be present — an empty scan set
# would pass vacuously and produce a green signal that proves nothing.

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# shellcheck source=lib.sh
source "$SCRIPT_DIR/lib.sh"

if [ -z "${GH_TOKEN:-}" ] && [ -z "${FIREWORKS_API_KEY:-}" ]; then
  die "redaction check has no secrets to scan for — refusing to report clean (set GH_TOKEN and/or FIREWORKS_API_KEY)"
fi

hits=0
scanned=()

# Normalized copy of the artifact tree: newlines and spaces stripped, so a value
# wrapped across lines by a log formatter still matches.
norm_dir="$(mktemp -d)"
trap 'rm -rf "$norm_dir"' EXIT
while IFS= read -r -d '' f; do
  case "$f" in
    *.zip|*.gz|*.tgz|*.tar|*.bz2|*.xz|*.7z)
      log "SECRET LEAK RISK: archive staged for upload cannot be scanned: $f"
      hits=$((hits + 1))
      continue
      ;;
  esac
  tr -d '\n\r \t' <"$f" >"$norm_dir/$(printf '%s' "$f" | md5sum | cut -d' ' -f1)" 2>/dev/null || true
done < <(find "$ARTIFACTS_DIR" -type f -print0)

# scan_value <label> <value> — checks the value and its common derived forms.
scan_value() {
  local label="$1" value="$2"
  [ -n "$value" ] || return 0
  scanned+=("$label")
  local -a forms=()
  forms+=("$value")
  forms+=("$(printf '%s' "$value" | base64 | tr -d '\n')")
  forms+=("$(printf 'x-access-token:%s' "$value" | base64 | tr -d '\n')")
  # URL-encoded form (only the characters that actually appear in tokens).
  forms+=("$(printf '%s' "$value" | sed 's|/|%2F|g; s|+|%2B|g; s|=|%3D|g')")
  local form found
  for form in "${forms[@]}"; do
    [ -n "$form" ] || continue
    found="$(grep -rlF -- "$form" "$ARTIFACTS_DIR" 2>/dev/null || true)"
    if [ -n "$found" ]; then
      log "SECRET LEAK: $label found in artifact file(s):"
      printf '%s\n' "$found" >&2
      hits=$((hits + 1))
    fi
    # Whitespace-stripped scan catches line-wrapped occurrences.
    found="$(grep -rlF -- "$(printf '%s' "$form" | tr -d '\n\r \t')" "$norm_dir" 2>/dev/null || true)"
    if [ -n "$found" ]; then
      log "SECRET LEAK: $label found (line-wrapped or whitespace-split) in a staged artifact"
      hits=$((hits + 1))
    fi
  done
}

scan_value "GH_TOKEN (App installation token)" "${GH_TOKEN:-}"
scan_value "FIREWORKS_API_KEY" "${FIREWORKS_API_KEY:-}"
scan_value "SHOC_REVIEW_APP_PRIVATE_KEY" "${SHOC_REVIEW_APP_PRIVATE_KEY:-}"

# Generic credential patterns: catches secrets belonging to the PRODUCT repos
# (a PR touching .env or appsettings) that the runner knows nothing about.
generic_hits="$(grep -rlEI \
  -e 'gh[pousr]_[A-Za-z0-9]{30,}' \
  -e 'github_pat_[A-Za-z0-9_]{30,}' \
  -e 'BEGIN [A-Z ]*PRIVATE KEY' \
  -e 'AKIA[0-9A-Z]{16}' \
  -e 'xox[baprs]-[A-Za-z0-9-]{10,}' \
  "$ARTIFACTS_DIR" 2>/dev/null || true)"
if [ -n "$generic_hits" ]; then
  log "SECRET LEAK: generic credential pattern found in artifact file(s):"
  printf '%s\n' "$generic_hits" >&2
  hits=$((hits + 1))
fi

if [ "$hits" -gt 0 ]; then
  die "artifact redaction check failed: $hits leak indicator(s) — artifacts will not be uploaded"
fi
log "artifact redaction check clean (scanned for: ${scanned[*]} + generic credential patterns)"